Skip to main content

Full text of "Isokinetic sampling of aerosols from tangential flow streams"

See other formats





R803692-01) froa tha Environaental Prntectign Agency (EPA), i 
■ortitored by EJA's Project Officer Kenneth T. Knapp. 

Inportant part that they played in ay edocaclon. I an eapocially 
appreciative of Dr. Umdgren for his guidance, oocouragCBant and 

preparing this oanuscript- 

Finaily, 1 wish to thank ay parents 

through the difficult tin 





AnlKakfjiatic SanpUnj 
Noisle Mlsaltgimenc 


UaloclCT DeteTnlnation. . . 

















projected area of saaplor inlet 

coQccncrction ratio of aoroaol generating aolution 
particle diarecer 

distance upstrean froni noazle 

stack velocity 



angle flf BiisaUgnosnt of nottln with rospecc to l 

A therefore el 

developed to edjust the Stahes nunber to eccount for i 
Using an aquation enpirleally developed fron the* 
using the equations of Belyeev and tevin describing aj 
bias with zero eiselignaent, a omchematical aodel was 
which predicts the sampHni error when both noszle alt 
iineCic sanpllnj velocities occur siaoltaneously. It was found that the 
saoplins bias approached a aariBuri error tl>ScosB| where R is the ratio c 
the free stream velocity to the sanpling velocity. During the testiug, i 
IS 60% of the particulate oatter entering the 

The causes and characteristics af tangential flaw streans ore described 
as they relate to probleos in aerosol sampling. The lialtations of the S-type 
pitot tube when used in a svirlinE flow are discussed. A three dinensionol or 
five-hole pitot tube was used to oap cross sectional and asial flow patterns 
In a stack followlni the outlet of a cyclone. Anglos as great as 7D degrees 
reletlve to the axis ef Che stack and n reverse flow core area were found in 

Using infonmcion found In this study, a sloulation oodei was developed 
to dccemiiie the errors involved when Baking a Method 5 analysis in a tan- 
gential flow streaa. For an aerosol with a 3.0 ua 1»ID [nass oean diaaeter) 

and goradlrie standard deviation ( 5 ^ of 2.13, the predicted eeracentratlon 

end a a of 3.3, a 20i error wns predicted. Flow races detenclnad by the 
S-c/pe pitot tube ware from 20 to 30% jrontor than the actual flow rate. 
ImplicacinDs of these results are described and racomMndntiona for aodifica- 

taniceotial flow stress ore 


This stud/ deals wi 
sanple of particulate matter from a gns stream chat does not flow 
parallel to the axis of the stack as in the mass of s-lrllns or 
tansentlal floe. This type of floe is eoimeonly found in stacks and 
could bn the source of subscantisl snmplini error. The causes and 
•s particular flow paccem are described and 

rate determinations are thoroushly anal/tod and discussed. 

The analysis of sajnpling errors is approached from two directions 
in this study. One approach involves an investigation of aerosol 
sampling bias due to anisokinetic sampling velocities and misalignment 
of the nozalo with respect to the flow atreatn as a function of particlo 
nod flow characteriitles- The second part of the study involves en 
accurote mapping of the flow patterns In a tangential flow aystem. 

The inforeatlon obtained in the two parts of the study will be combined 

Method 5 (i , 2] analysis in a tangential flow stream. 

tsoklnatlc SMpUnt 

To obrain a represontatiyo ftaople of poTtlCDlote Batter from a 
ooviri fluid. It is necessary to saaplc isokinetlcally. Tsakinetic 
seopliftg can be defined by too conditioner (3) 1) The suction or 

nosalo velocity, V^, oust bo equal to the free streasi velocity, V ; 
and 2) the noszle Bust be aligned parallel to the flow direction. 

onterlnj the nocile. {see Figure 1). Thus, there Kill bo no 
divergence of streamlines either away frcoe or into the nettle, and 

When divergence of streoBlinss is produced by superlsohinetic 
sampling, subisoklnotlc sanpling or noetic misBlignaient, there is a 
possibility of particle size fractionation due to the inertial 
properties of particles. In the cose of superisokinetic sampling 
[see Figure 2), the sompUng velocity, V,, is greater than the free 

sampied- A^', will be greater than the. frontal area of the sampling 

will enter Into the nettle. Particles ootsldo this area but within 

ic larger perticles will be ur 
bo sanpllng nettle. Since no 

velocity is less tine the free stress velocity Cso» Figure J) . In this 

plod area of the flov, The volusc of air lying aithin the projected 
area, Aj’, but outside aill not be saaipled and the streasllnos will 
diverge croued the nozzle. Hoaever, some of the particles in this area. 

of the particles within A , 

caused by snpsrisoblnetic saspling. When 

Therefore, whenever the noztle is misaligned, the 

if anisokinetic sampling [superisohinetic, 

concentration error will depend upon the sice of the particles. More 
specifically it will depend upon oarticle inertia, sdiich iaplies that 
the velocity and density of the portlcte arc also inrportimt. Portlcle 

inert!* affetts the ability of the particle to negotiate turns with 
its streenline which deCemlnes the seiount of error. Therefore, in 
ail cases greater saiapUni errors will occur for larger particles an4 

Besides dotervining the direction of the sampling bias, it is 
also possible to predict theoretically the ninimuB and noiioan error 
for a given condition. This con be done by considering whnt hsppens 
when the Inertia of the partlclei ia very small (i.e.. the particles 
can negotiate any turn that tha stmaalines mabe] and what happens 
when the inertia of panicles is very Itrge (I.e., the panicles are 
unable to negotiate any turn with the streaiallnes) . In the former 
case of very low inertia, it can easily be seen that since the particles 
ore very mobile they do nut lesve their streamlines and therefore there 
will be no sampling bias. In this situation the concentration of 

sampling depends oi 

ye accurately obtained regardless of sampling 

■ 0 is obtained for amall inertia panlcles- 
:hat can theoretically occur in anisobinetic 

In the case of unequal velocities for very high inertia particles 
*Aich are unable to negotiate any change of diroclion, only those 
particles directly in front of the projected area of the noillo, Aj. 
will enter the nostle regardless of the sampling velocity. Therefore, 


le litei-iturg on Aplaokinelic Sfplin | 

Sa»pllm 8ia» Du» to UnnitchtJ Vi 

Numerous eiticles hsve been urltten destriblnj tHe sources end 
rapilcudc of errors when isokinetic conditions are not sslncained. 

In one of the ceriler uorks, Lnpplo end Shepherd (4) studied the 
trajectories of particles In s flov stroajn and presented a fonmila 
for estiMtIoj the order of the magnitude of errors resulting alien 
there is a difference boteeen the average sampling veiocity and the 
local free stronn velocity, hatsem CS) examined errors in the anlso- 

diameter (iKO) and found the relationships shown In Pigure 5. Super- 
isokinetic sampling (aampllng with nottlo velocity greater than the 
free stream velocity] leads te a concentration less than the aetuai 
concentration, while subisoklnetlc sampling hos the opposite effect, 
tfatson found that the magnitude of the error was not only s function 
of particle site as seen In Pigure 5, but also of the voioeity and the 
notile diameter. He propoeed that the sampling efficiency was a function 
of the dimensionless particle inertial parameter K (Stokes numher) 
defined as 

; « Cunninshnm eorroctlon for slippvge 

r - PpOp'/Un 
1 - viscosity of gas 

correct tvichin IQi, the velocity ratio R «ust lie becveen 0.R6 end 1-13 

datcctablo concentration changes even while saepling at a 4003 variation 

relatively onlaportant for fine particles. Heneon and Haines (7} 

site ranges (5-2S, 80-100, and dOO-500 pm) and in a range of norslo to 
staefc velocities of 0-2 to 2-0- They found that where the velocity 

proxinatciy 803, and that deficient noaile velocities resulted In greater 

for the coarse particles, the velocity into the noaale had no important 

besrins on the quantity of dust collected. They suggested 

produce of the nostle area and the stack gas velocity approaching the 
nettle as tho gas senplo volimo, regardless of Che velocity of the 

Is possible to obtain small deviations even where departure from 

sampling anisoklnoticany For large particles, 

Lundgren and Calvert (9] found the sampling bias or aspiration 
coefficient A, to be a function of the inertial impaecicm parameter K 
end the velocity ratio R. They developed e chart which can be used 
to predict inlet anisokinetic sampling bias depending on both K end R. 
Redtloch's (lOJ oquacions defined the dependence of the efficiency upon 
particle inertia and the velocity retio. In a allghtly different 

A = C./C^ " I • (R-1) 8(K1 (S) 

where B[K} Is a function of inertia given by 

8tH • [ 1 -exp {-L/t)]/(L/« C6) 

g Is the stopping distance or the discanco a particle with initial 
is defined by ( 11 ) 

L is the dlstence upstream from the nozsle where the flow 
by Che downstreen nozzle. It is e function of the nozzle 
is ^Iven by the equation; 

is undisturbed 
diameter and 

hem to verify Badzioch's claim that L, the undisturbed distsnee 

anisokinetlc sampling 

F previous studies on or 


:t slsnificant changes li 

Figure 6 ahovs a plot of eqoations (5] , 

BOFonB R a 1, Che aeplration coefficient cends co as9ympcaciceU>r ap- 
proach its cheorecical lieit of R. Reyoml a Scokoe number of about 

il consideretlona. Bodtioch (IB} and Belyaev and Lovln [12) 

IB screajDiines a 

art to diverge at approxioetely 6 
directions in an anount of tine 

the linlclng sice particle 

ttialyzins concentration errors olotsined while sateplinf suboiicron 
particles, O.S ub XhS> and 1.26 jeotiietric standard deviation, traveling 

sanpllng velocity wos 206 oP tho free screen velocity (ReSl- 

field was eisintaincd or assuned c 

probe niaalignnent do not provide enough puantilntive 

reported by Ifatson (61. on Che effect of eisaligntwnt 
efficiency of 4, 12 end 32 Oat particles (see figure 7). In a : 

lecced the highest concentration. Although 

Figure 7. Error due to aiselignioent of probe to flow 
Hayhood and Langstrotb, in Katson {E)J. 

theoretical predicciotis fi.e., aeasured conceatraTion 1; 

inportanc paroaeters, free atrean velocity at 
lot included in the analysis. 

»lod speeds of 100, 200, 400 and 700 ca/sec vith the notslo aliened 
over a rnnje of ongles froo 60 to 120 dojrees- He then used a trijono- 

a • 1 • e[«[(VjSlne 

V^CO!B)/IV|C05« • V^sinO) - 1] (11) 

serves to invert the velocity ratio betveen 0 and 
I not realiscically represent the physical properties 
In tact, equation ClI) hecooes unity at 45 defroos 
the velocity ratio or particle site is. This cannot 

1 decroase inversely proportional to the angle and 

A more representative function can be derived in the following 
velocity V^, let ho the cross sectional area of the nettle of dianetcr 

0 in ecoistiem (U) to r< 

iB the Jiniting situation for a pnrcieie to 

proportional to the avcroge OiaButer of the frontal area of the not) 
Puchs [19) suggests that for srsoll angles the sampling efficiency wl 

Laktionov [20} sampled a pel 

used a photoelectric instellatlon 
cion coefficients for different s 
ever a range of Seohes nunbers fri 

three subisoklnctic conditions. He 

:ed particles. Pron data obtained 
1 0.003 to 0.2 he developed the fol- 

A few w»lni=al studies in this .rea hate also been published. 
Dasio!’ ( 14 ) theoretical calculations of pottleU trajectories in a 
nonrlscous flow into a point sinh deterained the senpling accuracr 
to be a function of the oostla inlet orientation and dia.eter, the 
sasiplinB flow rote and the dust particle inertia. Vitols (Jl) also 
•ado theoretical oitlnates of errors due to onlsoklnctU sanpllni. 

Ho used a procedure conbininB an analog and a digital coaputer end 
considered inertia as the predoininent ■echnnism in the collection of 
the particulate itatter. Howocer. the results obtained by Vitols are 
only for high values of Stokes numbers and am of little value for 

»^_Sierory of the lltarcture en Tinientlal Flow 

;c aampling velocity is known ti 

particle saapling bi 

(e.g., centrifugal, electrical, gravltetlonal or therMlj; and probe 
mlsollgiuaenc due to tangeatlal or circulation flow. These factors ai 
al«n always present in an Induscrial stack gas and cannot be assuiu 
to be negligible. Not only do these factors cause sinpling error 
directly but in addition, they cause particulate ct 
and aerosol size distrlhution varintlons to exist a 

Tangential flow is tho non-randotn flow in a direction other than 
parallol to the duct center llua direction. In an air pollution 

control device, Hhcmever centrifugal force is used ns 
particle collecting aeehanism. ungentiel flow vili oecur. Css 
flortng fron the outlet of a cjntlono is a classic erasipU of tongeaciol 
flov and a well recognised probleo area for accurate particulate sampling. 
Tangential flow can also b. caused ty flow ch.nges induced by ducting 

tangentially, a 

T Csee Figure 9), Ever 

within an order of oagnitude of tho stack flow 
flow pnttom will occur Csec Figure 10). 

The swirling flow in the stack combines tl 
vortea laoticFn with asisl motion along the stack avis The gas stream 
moves in spiral ar helical pacha up tha stack. Since this reproscnts 
a developing flow field, the swirl level decays and the velocity pro- 
flies and static pressure distributions change with niial position 

nous ere composed of aiiol, radisl and tangential 
escabl ished 

tangential or 

or circumferential velocity components Cs«e Figure JI). 

vortor flows ore generally amisyeanccric but during formation of t: 
spiraling flow tho s.vmmetry is often dietorted. The relnllve ord. 
magnitude of the velocity components veries across the flow fiold 
the poasihllity of each one of the cottponents becoming dominanl ai 


The two distinctly difforont types of flow that are possible in 

When the swirling cospcmedt of flow is first created in the cycloee 
exit, the tangential profile of the Induced flow approaches that of a 

by dividing it 

is loss of angular ■omentum is due to viscous action 

angular noniontuia and tengentiai 

1 and tangential point volocity 

■ tangential velocity profiles and 
la distributions are plotted in Figures 12 and 13 fron 
lien at 9. 24 and 44 diameters dawnsciesD of the origin 
'low. The tangential velocity (VO is made diaensionless 
by the mean spatial axial velocity CU,) at a pipe cross 

developacnt i» due prinsrily to yieeoslt/ e1 
vary dependent on the in 

suirle may produce reversed axial velocities in the central region (21), 
It should be noted that although tangential velocities and angular 

after S4 diameter the tangential velocity is still quite significant 
vhen eoapared to the axial velocity. Therefore, satisfying the GPA 

the tangential component of velocity was as high as SO degrees at some 

m gradients and invalid flow 

described In the previous chapter. Concentration sradients occur 
because the recatitmsL flOM in Che stoc): acts aoaeuhac as a cycione. 
The centrifugal force causes the larger particles to move touotd the 

to deteteine the teagnlcude of th 
cyclonic floe. Results of floe 

ot of a small industrial cyclone 
determined at the different 

errors can result in cases of tangential flm. 

eleiost 741, Hhen sanpling domstreosi of the 
however, the error was reduced to 15^. 




Saapllnjt at the anfle of Baainun velocity head i 
to The reaults cannot be coapared directly tc 

parallel aanpling approach because the feed rates were not the scuiie 
duo to equlpnent failure end repUcetnont- Sasipling in the straighconed 

pacted on the etraightening vanes and settled io the hcrritootal section 
of the duct, thus ressoving theta froia the floe streaei. 

Particle siae distribution tests showed no significant effect of 

portieles being too small to be affected by the centrifugal force field 
sot up by the rotating flow. 


Aclusl Ejoission 
Hate tgr/js en 


sijice It has larga dioaeter pressure part 

that uill not plu| (see 
rts It has on additional 

SQBevhat insensitive to i 
and 16 shoe the veioeity 
5-type pitot tube is vorj 

ihe for a given velocity. Houever, although 
I give an accurate velocity aeasurement. It is 

irrors for yaw and pitch angles. Although the 

luse of this Insensitivity to direction of flow 
S-type pitot tube cannot bo used in a tanEontlal 

The Bagnitude of the radial and tangential conponents relotive to the 
axial component will doteminn the degree of error Induced by the tangential 

the flow rate through the stack, but both affect the velocity neasureitcnt 
inado by the S-type pitot tube because it lacks directional sensitivity. 

If the maxljfruiB velocity head were used to calculate the stack velocity, 


tube in cingontinl flo» 

1- Methods Jtvellible fc 

PToi Pielj ~ 

M neither the rsdlel velocity, V . th 
irial velocity, V^, nor the enjle « ct 

a tenjeotial flo» field hive been hised upon introduction of probes Into 
the flou. Becnuso of the sensitivity of vortor flo.s te the Introduction 

Tvo comnion types of pressure probes capible of aeosuring velocity 

>c dinenstoncl directional 

a henisphericnl or conical probe tip 
one on the axis and at the pole of the tip, the other four spsced 
equidistant from the first and from each other at on angle of 30 to SC 
degrees frets the pole. The operation of the probe is based upon the 
surface pressure distribution around the probe tip. If the ptobe U 

then a prassure differential vill be set up aerreas these holes: the 
inasnltude of uhleh «ill depend upon the geometry of the probe tip, 
relative position of the boles and the magnitude 

differencials becveen holes as a fuoccion of yaw and pitch anples. 

Figure Id shows the sensitivity of a ty'pical S-hole pitot tube to ynw 
angle. Because of its sensitivity to yaw angle, it is possible to rotate 
the probe until the yaw pressures are oijnal, ueasuro the angle of probe 
rotation tyaw anglel and then determine the pitch angle from the re- 
maining pressure differentials. The probe cin be used without rotation 
by osing the cosplete set of oalibrotion curves but the ceaplexity of 
■eesureaent end calculetion is increased and accuracy is reduced. Vel- 
ocity coiponents can then be cslculated froo the measured total pressure, 
stetic pressure and yaw and pitch angle aeasurements. 

unable to measure pitch angle. The probe is character! tod by a central 
total pressure opening ol the tip of the probe with two static pressure 

degrees. From Figure IS it con be soon that the probe is quite sensitiv 
to yaw angle and can therefore be used to determine the yaw angle by 
rotating the probe until the pressure readings at the static taps are 

S. EPA Criterii fet Saapliog Cyclonic Flew 

The revisions to reference methods 1-g (2) describe a tost for 
deteminneion of whether cyclonic flow exists in e stock. The S-type 

Figure IB. rccHctmer pitot luhp ronttltlvlcy to >-ow angle. C2B) 

m relative to the 
le ^essure reading 

je ij used to detenine the angle of the f] 

■ho stack by turrdni the pitot tube until i 

id of Method 5 should he used to ssnple the 
'o procedures include iestallotion of 
straightening vanes, celcoiotinj the total voliaietric flou rate 
stolciioeiotrloally, or moving to another masauresent site at vhich the 

Straightening vanes have shonn the capability of reducing suirllng 

the physical Unltation of placing then in an eslsting siact. Another 
ia the cost In terns of energy due to the leas of velocity pressure 
vhen ollnilnstini the tongcetlel and ridlol conrwnonts of velocity. 

Since the vorteg flows are so sensitive to downstroon disturbances, 

It is quite possible Chet straightening vonee might have e drastic 
effect on the performance of tho upstream cyclonic control device 
which is genemting the tangential flow. Because of these reasons the 
use of straightening vsjies is unacceptahle In many situations. 

Calculotlng the volumetric flow rate stoichlomotrioally might 

CBlculatc tho necessary isokinetic samnling velocities and directions. 

Also, studios renortod hero have shown that the decoy of the tangontiol 
eoopooent of velocity in circular atocks is rather slow end therefore 
it would be unlikely that another mensurenent site «old solve the problem. 

as obsorvlng tl 

8 BtscV is correct. Other approaches such 
n of the plume afeer leevlng the stack could 

I [28) f, 

twin-spiralioa vorticlos often seen leaving stacks 
secondary flow effects geoerated by the bending of 

A. Expcrtfttflntat Design 

Pigure 20- An aerosol stream ganeriled Prom a spinning disc genamtor 
was fed into a nlxinf chojiiber xhera it wos coabinad »ith dilution alt. 
The air stream then flowed through a 10 em dlamoter PVC pipe containing 
straighconing vanes. This was followed by a straight section of clear 

as a control sample originated In a box follOHing the straight seetlen. 
A test noaala was Inserted into the duct at an angle from outside the 
box, A thin-plate orifice, uaod to monitor flow rate, followed the 

through the systaa. The flow t 
dieieetor of an orifice plate. An air by-pass between the blower and 
the orifice plate wea need as a fine adjnst for the flow. 

The aompUng systow (sou Figure llj consisted of stainless steel, 
:ted to d 7 am stainless steel Gelman filter 
Each filter asaenbly was connected in scries to a dry gas 
ind 0 rotameter, end driven by an airtight pump with a by-pass 
.0 control flow. 

by changing tl 



MOToaispSTso o»ro5»u froo 1-0 NMD to 11. 1 » NMD (seo Tatu IVJ . 
Droplota «org generated fro* a sLiture of 5D% uranine (a fliioresoent 

ethanol (9S» pure) and up to lOt distilled/deioniied lljO. Uranine »os 
uaed ao that the particlea could be dotocred by fluoreaetrie aethods. 
Methylene blue xaa added to aid in the optical stiiej of the particlea. 
The Birture of water end ethanol allowed for a unifora avaporotion of the 
dropleca. The droplata, containing dissolved aoluto, evaporated to yield 
particlea whose dianeters could be ealculated from the equation 

D = particle diaaeter, uti 

■ ratio of solute volune to solvent volutae plus solute 
voluae, diaensionless 
Dq ■ original droplet diametar, |aa 

With Iho diac'B rotationel velocity, air flows and liquid feed rate holt 
constant the site of the droplets produced were only dependant upon tho 

a dynamic force balance between the centrifegal force end the surface 


Ethsnol Diameter, uj 

Spinnins Disc 

Spinning iliac 

Spiening Disc 
Spinning Diic 
Spinning Dlac 

Spinning Disc 
Spinning Disc 
Spinning Disc 
Spinning Disc 

ro respectively 72.5 anrt 22.3 dynes/ctn f3 

C. Velocity Deterainition 

The velocity el eech sonpllnj point ns lOMSured using a standard 

trolling the pressure drop across a thin-walled orifice placed in the 
system [35-37]. Five orifice places with orifices ranging in disneter 

A typical velocity profile across the 9.5 cat clear plastic duct 
is presentod in Table VI and plotted in Figure 22. The profile la 

Reynolds m 

yr this particular case was 1.1 x 10^. The velocities 
of isokinetic sampling rats and StoVss nuinbar. The difference between 
plate calibration is probably due to the inability of Che pitot tube to 

existed across the diameter of a cylindrical duct, and that the 
magnitude of Che concentration gradient varied with particle size. 

le noztles were made approximately IS cm 
10 dlsturfaanoe caused by tho filter 


ne«s, the releUve ectge thickness 

F, Anslysls ^ecedure 

Uranine particles were collected on Gelnurn t/pe A floss fiber 
filters. The filters vere then placed in s 250 ml boaher. One hundred 
sllllliters of distilled wsur vere then pipetted into the front half of 

analysed by a fluoroneter (39] . 

le particles were 5.0 ub type SH Millipore membrenc filter.s. In ardei 

therefore, each filter wes dyed with inh and a grid ue$ draws to aid 
in the counting. Sefore being placed in the filter holdere, the filters 

with black gride. The Isopropyl 
reiBovo background particulate aai 
the alcohol was poured into the 1 
through the notrlea. The solutii 
filters. The filters were allowed to dry and 

filter holder wan analyiod 
ce the background wan low enough. 

Saepline Procedure 

Ig the by-past as a fine ad}uat. 

'as ncnsured using a standard pitot tube, 
sit solution was selected for a given particle 

uslag n light microscope. 

saopligg race clasoet to 1 
Isokinetic snnpllng rates 
flow rates were adjusted accordingly- 

angle- SanpUng tines varied froo IQ to 20 nlnutes. 

The syeten used tc 

tr, follawed by e 6.1 meter length ef 2G cm PVC 

Pitch angle is then detenined 


boajuro ae the flow perpendicular to the axis of the stack and tangent 
to the stack walls. The pitch angle is o neasuTe of the flow perpendic 
ular to the axis of the stack and perpendicular to the stack walls. Th 

where = eooponent of velocity flowing porallel to the axis of the 

[cosCpitch) X cosCyaw)) 


A. Aerpsol Sampling EKperiwants 

EApcriaencs wnra sat up and run with StoVos number ss the Independent 
variable. Duct velocity, notsle diameter and particle diameter were varied 

Stokes number used in the analysis of data was calculated from 

density of ragweed pollen ° 1.1 g/cn^ tlB} 

DO ac both acmpUng locations, sianilcaneoua saoples 

This can be aapected because a staall error in jrrobe nlsalUnnenc would 
have a greater effect at the higher Stokes nuaber. 

In the enalrsis of the tests using ragweed polien, the filter 

particle, analysis of the probe wash Is a necessity. An average of 

parallel to the flow stream and sanpling isokinetioally. Therefore, 
the loss of particles was due to turbuleat deposition and possibly 
bounce off the filter, and probably not inertial Inpoction, For tests 

impnetion of particles 


degrees (see 

The probe «»sh for eljht tests using 4.7 u> uranlnc particles 
was also ajialyted separately for comparison with the results of the 
ragweed pollen tests- While parallel sampling, from IS to 3Sb of the 

holder, while this was somewhat less than the amount of rogweed pol- 

variatlon of the percent collected in the nozile during identicnl tests, 
the probe wash cannot be accounted for by n correction factor. During 
further testing, it wes qualitatively observed that the percent in the 
probe wash increased with particle site and decreased with increasing 
nottle dlaaetor. 

t. The Effect cf Angle aitsel Itoment on Sempllne Efficiency 

The aspiration coefficient wes determined by comparing the amount 
of particulate matter captured while sampling isohinetioany with a 
control noasle pieced parellel and a tost noaslo set at an angle to tl 

For all three angles the aspiration coefficient approached I for smaU 
minimum cf cosfl for largo values of Stokes number. The most 


IS rspresent the sejnpHng efficiency as a 

randoiBl/. This was done to check the lojitljiTacy of uslo| Stokes nurobor 

dianecer is Laportant because it deteimines Che amount of time available 
areo and cherofOTO the projected notzle diameter ore reduced proportiooel 

d r 


coefficients for 30, 60 and 90 degrees should approach tbelr theoretical 

aniscklnecic sampling velocities (see Figure 6). 

To develop the adjustment factor for Stokes number. It was neeos- 

O.S. Therefore the value of k of Interest is where there is (,9Sj(0.51 - 

not possible to detect esaetly when the curve reached 9Sk of its niniinum 
value. Therefore no value for 30 degrees was used In this anslysl.s. 

The equation for the adjusted Stokes number determined from Figure 

be imjltiplied by 1.93, 

While ectetnpting to determine the constants a and b, it wae found 
that the fore of the eouation had to be altered swnewHat to allow fl' to 
approach 1 at a faster rate for ualuos of h’ greater than 4.0. The fol- 
lowing is the flnoZ form of the equation selected. 

sempling efficiency due to noitlc misalignment as 

IS nuabor. An oxamplc of this 

dotroBS. Iti order to ssmplB isoklnccicslly such thot thcro 

is no dlvergencB of streamlines into the noxslo, the sample velocity 

the condition of H e lynose defines Che condition for obtaining a 
reprosnntative saaple xhen the nosiie is misalignod «ith the flov 

Since Che snnpilng acthodologies used to determine fiCS<h} 
fi’(K,9J were substantially different (photoiraphic ol 

sampled isokinotically and the tost notile sampled anisoklnctlcally. 
Tests uere porfomied at tvo Stokes numbers (k = 0.1S4 and k • 0.70) 
and at tuo velocity ratios (B = 2.3 and R = 0,51). The aspiration 
coefficients obtained by comparing the tvo measured concentrations are 
presented in Figure 33 and Table XIII. The data obtained lie vithln 

data [Equations (5), (9) and (10)|. 

Since tho two methods give conparable results, experimoncs wore 

flow streoB end the sampling velocity was set to be isokinetic. The 


in Tobies XIV and XV, Thosodaca ore plotted and conpared with the aodel's 
prediction in Plguto 34. The aspiration coefficient does indeed appear 
to bo unity uhen B = l/cos8 as in the ease of B = 2 and « • eo degrees. 

limit of RCO50 CO. 25) at approximately a value of Stokes number of 2 to a 

degrees, R b i approaches its theoretical Unit. This further confirms 
the necessity of using an adjusted Stohes number vhen the probe is mis- 
aligned with the flow stream. 

To ftkrther test the nodal, experiments were run at 45 degrees 
in Tables XVI and X\1I arc plotted in Figures 55 and 56 also show good 
When tests were run at 9 s 90 degrees, R = 2,1 and K = 0.195 Csee 

Table XVIII) , an average aspiration coefficient of only l.Si was obtained 
The value predicted for equation 131) for these conditions is 49t. it 

with ths prediction nodel, they do compare favorably Kith the ettpericol 
equation of Lahtionov {20] [equation (16]]. For the conditiona of r » 

feet that two completely different sampling schemas were used, and 

. Tansantial FI 

. H'a’i”! 

occording to EPA Method 1 (1) [see Table SIX). Messuremonts vere made 
each point in the traverse, the pitot tube uns rotated until the pres- 
froB ail five pressure taps were recorded for later cslculation of total 

During the Initial velocity tt 

the flam could not be 
IS charactoriaed by 


Tobies XX-XXIV shOM Che colculaced results of the eeloclty mcesure- 
nients Bt the five axiol positions. The low flow was the flow Boasurecl 

when e Tcstrictlon was placed at the 
Induced Bpproxioacely a 40\ decrease 
represented o voluaietric flow rate of 

^ 5k sir 3S£ 


point to thff opening, one ol 

\e pitch pressure poin 

if the dste. 

The velocity- in. 

flov rates and all five aaial distances showed approsimately the sane 
characteristics. The pitch angle increased freo the core area to the 

core area to die walls. At the inlet and up to eight dianetors dtnm- 
strenm, aoglos as high os 70 degrees were found near the core area of 
Che flow field. The total velocity, aaial velocity, and the tangential 
velocity all showed the sane cross sectional flow pattom- The velocities 
were ninlnun ot the core, increased with radius and then slightly dccrcasod 
near the wall. These patterns arc similar to those found in the swirling 
flow generated with fixed vanes [23) . 

In otdsr to observa the changes in the flow as a function of axial 
distance fro. the Inlot, Che cross seotloa.I averages of the angle *, 
core ares, sad tangential velocity were calculated and presented in Table 
XXV end plotted In figures 35 and 39. All three parameters show e very 
gradual decay of the indicators of tangential flow as was expected from 

s [23). 

IS confimod by repeated 



Avoraje Vnlutj for Hi|h Floi 

Diajn«tn'5 4 

Ektwnstraajn fEegrccsl 

Location of* Tongentisl Velocity Cote Area 
Core Atea fowl [co/sec] fca’] 

Average Values For Low Floe 

velocity nt 2 diajnotors fron 

.n Hvernge tangential 

!r velocity 

seen fren the other profiles tt 
bisber velocity average. 

Plotted in Figure 40 is the location of the core area with 

avisyannecric and the location of the core area changes It 
axial distance. Only one drawing is used to ropresenc the situation ft 
both high and low flow rate because the location for both conditions wi 



A MNlel h05 bAsn dbvolopsd snd test«d wlilch doscribes particle 
collection efficiency as a function of particle characteristics, angle 
of nisalignaenc, and yoloclcy ratio. Together Hith Che neasuresent of 
velocity components in a svirling flow it is possible to analyte the 

analysis of the effluent stream folloviitg a cyclone. 

For this slnolation enalysis, the volumetric flau rate end iso- 
hinecic sampling velocities aro cslculnted from velocity aeosuremencs 

obtained at the eight diameter saagiling location using a S-type pitot 
tube (soo Tables XICVI and XNVII) , The angle d, velocity ratio, and 
particle velocity aro deiettiined from velocity oioasuroiaonts made at 
tho same location using the five-hols pitot tube (soo Toblo Mill- 
The particle characteristics are obtained from particle size discribotion 
tests node by Mason (22) on basically the same system. From a particle 
distribution »ith a 3.0 ua reiD and geometric standard devistion of 2.13 
(see Figure Cl), ten particle diameters were selected which represent 
the midpoints of IDi of the mass of tho aerosol (see Table mill). Tho 
density of the particles was assumed to be 2.7 g/cm^. The nozzle diameter 
was selected using the standard criteria to be 0.63s cm (1/4 inch) . In 
the model it wns assumed that the nozzle would bo aligned parallel with 



equations (33) . (39), 

the average aspiration coefficients are 

A,(Rj,ej,iCj) . ,lys^ - . .IA„p,5, * ... * ,lAj,p55, {33) 

= -IAq^, * -lAjjpjj^ ♦ -‘''upjsA * ••• • -“dpssa (Si) 


s' ’ ■'‘'npsi * '**npi53 * •'''rpjst ' ••• * 

ol voloclti- at IVSaopIin* velocit)’ at i) . 

'‘ili* StoXes nuober based on the nozzle diatoecer, total velocity 
at 1 , and particle diaBeeer PPjjg. 

Dpjj- Midpoint particle diaaeters each ropreaentlnj IM of the 

Since the saaplloj velocity »ill deteraioe the volumo 
oocli traverse point, the total aspiration coefficient 
as aletermined by taXlng an average ueiphted according 


Khore = iniet velocity at traverse point j. 

Bocoose of the Hissing data at point I and negative pressure settlon 
point . 1 , these t« traverse points veto 


The totil aspirecion coefficients 

Table XXIX and XXX). There are two reasons for the relattye lo« amoim 
of concentration error found in this anaiyila. One reason is that the 
two nechnoisns causinj sopapUnj error, nocile i.isoli)!niBotit and aniso- 
kinetic aanpllng velocities, cause errors in the opposite direction. 
The S'type pitot tube detected a velocity less than or equal to the 
actual velocity which would load to oublsokinotte sanpllnj producinj 
an increased concentration. The notsle ■Isnligmnent when sanplini 
parallel to the stack wall would produce a decreased concentration. Sc 

values lead to sisall snapllni errors, even when isokinetic sanpllni 
conditions are not oaintainad. 

Mason erperlnentolly detnmined that the collection efficiency she 

prcainneely nidway between the hixh and low flow rate in this studv the 
flow pattema should be approxlwtely the same. The dlicreoancy between 
Mason's erpnrinental values and the values predicted by the simolatinn 
probably tan be accounted for as esperliaontal error by Mason. It would 
be nearly impossible to obtain a SOX sampling error for an aerosol as 
small ns the one used without eatramc anlsohinetic sampling conditions. 

Senplinit Velocity 
from S-Type Pitot 
Tubg f^secl 

Ntighted Average 

the naxltein 4)1- From Figure 15 It is apparent that by sptlttini the 
difference hetueen the angles where the velocity pressure drops off 
rapidly, it should be possible to gee within 20 degrees of the tero 
yaw nngle. This means that the sample velocity aeasored by the S-type 
pitot tube will be approxijiiBtely the same as the true total velocity 

In order to see how much greater the error would he for larger 
particles, a siellar analysis was perfomod using a dietributton with 
a 10 urn mass mean diameter and 2.3 geueetrlc standard deviation [see 
Figure 411. This was the distribution obtained at the outlet of .i 
eyelone In e liol-mia asphalt plant [43). Because of the larger dlametor 
particles tlie saapling efficiency ves reduced to 0.799 for the high flow 

five-hole pitot tube dnte an-celculoiad by multiplying the average e« 
velocity by the Inner duct area minus the core area. The flow rotes 

the average «ial velocity. The voluaetric floe 
•Jltiplying the average axial velocity by 7/8th o 

nation of the average velocity and the entire inn 

The results presented In Table 71tX. shoe that the insensitivity oi 
he S-typa pitot tube to yae angle produces a higher calculated flou 

le avernge velocity detontlnntion, this error Is reduced to 174. 

luid be expected Trorn looking 
> you angle {Figure IS). When the trovci 
ah appreximately 45 degrees, the S-type 
ose to the total velocity. Hovever, bej 
oc tube readings drop off quite rapidly 

ibe readings vere 
'0 degrees, the pito 

are presented It 


u bsttor undbrscandlnj; 

obtain a roprcsentativa sanpu of particulate omcter ftoa gas sttoaiia with 
coDploa flow pattrma. Tbe errors ioduced bv tangential flow were analyted 

relative to the flow streao velocity, and angle of the nottle relative to tl 
direction of flow. The second involved analysis of swirling flow patterns o 
tjielr eubsopuent effect on flow teeBsuroaents made by the S-type pitot tube. 

■IsallgniaonT wore studied by taking conparativo anisokioetic and iaokinetic 
saoplea frm a straight section of duct. By anolyting the problan in this 
■ethod the data obtained are more useful and have many sore applications 
beyond this study. They provide fundamental inforaation for a better under- 

ilonal profiles were oeesured at five asiel distances along the stack to 

Ths two aspects of this study, anisokinecic soapling BTTora and 
flow iDoasurooents, were coablned in a slnulation oodcl to dcceriBine the 
mainitude of errors whan an EP4 Method S eolssion test is perforoed at 

A sumary of tho laportoat results deteinined froo this study is 
as foUoMsi 

Tho flow patterns fc 

'0 degrees relative li 

such as swirling flow, is inherently self'presorving in round duets, it 
decays very slowly as it laoves up the stack and therefore sanplins 
any location downstreaa of the cyclone will Involve the sane problem. 

B. The yaw characteristics of the S-tj-pe pitot tube load to several 

equal to tho actual velocity with the nayiautn error being appronioately 
5t. Beyond 45 degrees the aeasured velocity drops oft quite rapidly and 
at on angle of 70 degrees the aeasured volocitv is loss than half tho 

true velocity. Beesuse of its yaw characteristics, tho S-type pitot tube 
is not suitable for distinguishing the aslol oooponent of flow from the 
total flow which includes the taagential component. Voiumetrie flow 
calcuintlons bused on S-type pitot tube neasureeients in a swirling How 

tot tuoes boseo on cho fiye-ftole nnd throe-hole designs am 
s in detoTBining the velocity conponents iji a tangential 
The five-hcio pitot tube has the advantage of giving pitch 
1 yen angle. Hovever, In a cyclonic tlo« 
iroaia, the yaw angio Is of wjch greater tnagnitude than the pilch angle 
Id therefore, the pitch angle can be Ignored with small error. In the 
.tuation modeled, if pitch angle were ignored 

Informetion ns 

D. The particle aampling errors due to nnlsoklnetlc sampling 
velocity and notzle aisaUinment were analyied and a eodel vas developed 
to describe the saapling efficiency as a function of velocity ratio (R} , 
misalignment angle fS) , particle diateeter, particle velocitv, and nozzle 
diameter. It was found that the maaiaum error far R » 1. approached 
(I - cosS], Mien both a nozzle nisalignBent and anisokinotic sampling 
velocltie.s are involved then the maslnim error epproachea |i - RcosSl. 
The aquations and their limiting conditions for predicting the aspiration 
coefficient are summarized in Table gXXIl. 

sties of particle aampling. However, when tho 

hlch is due to a reduced pro;ectcd notzle dine 
IS developed to ndjuBt cbe Stokes number to to 

IS analysed separately from tl 
Ik of tho total particulate at 

9 has iDpUcationz 

Jut Borc laportaatlj' It Impilea that there nay he poasihu i>rohI»m 5 
m obtainlpj auurste particle site data nsinj a device such as an 
•tapactoT. If the ccllectlon of particles in the pottle is particle 
^iio dependent, then losses in the probe could lead to particle 

G. A similation nodel vas developed nhlch incorporates the 
inforsotion obtained in this study on particle soBpllng errors and 
the flou napping data. The particle sampling efficiency in a 
tangential flov stream ns, as eapectad, a function of partielo site. 
For a particle distribution with s mass scan diameter (IKD) of 3.0 
pm and a geanetric standard deviation of 2.13, the sampling errors 
predicted more less chan 104. For a larger distribution with a mass 
mean diameter of 10.0 t* and geoiaetrie standard deviation of 2,3, a 
304 sampling orrar uaa predictad. One of the reasons that the saitpllag 
errors were ss small os these ware, is that the two mechanisms inducing 
sampling bias produce errors In opposite directions. The missllgnitont 
of the nottle caused by the tangential velacicy component leads to a 
reduction of sample concentration. The reduced sampling velocity, 
calculated from 5-type pitot tube meesuremonts, leads to lublsolcinotlc 
sampling and an increased seaple ci 

if the ivernEe anjie of the flo» relntlve to 
the aiii of the otaet ij groacor then 10 ieirees, then EPA Method 5 
should not be pstforaed. Sinoe the luslniua error io particle soaiplinj 
hes been found to bo (1 - RcosOi , the 10 degroe requirement Is undul>* 
restrictive end a 20 degree linitotlon .ould be more opproprioto. Per 
a 20 deiree angle, the velocity neasured by the S-typo pitot cube uould 
be epproaimetely the saiee ea the true velocity Cl.e., a . I). Therefore, 

■e tl • CO 

ir moving to another location, because 
these suggestions, a better approach 
that it could bo used in a tangential 

either straightening the flow 
of the physioai limitations ol 
would be to laodify Method 5 si 

pitot tube, the direction of the flow could be accurately detorntinod 
for aligning the nottle, end the velocity components could bo measored 

threo-holc pitot tuba, the modification would hovo to inclndo a pro- 
tractor to measure the flow angle, an estra Tnommseter. andamothod of 
rotating the probe without rotating the entire impinger box. 

S5Sn;.:; ’ 

■ ■' 

■ tsrii a:.r WJ»«aR SK- 

■ s a.K?".::.:~'£=S--a 

■ src.’a**.'£r -ass.iTSi “ 

■ :;:Si LpSKS." ss.a',s! :s“'x,s.”5s.-' 

“■ :“-t? srr» 

'•■ »y.»3:::;‘:.!: sr;=,‘fsar sa.'Si-C'aaa* 

"■ as= uaa\.arn;^sa=;.t;‘£j;:,'E;.. 
■■- sa£.a£srs-,;:.£,‘ss,E“;,“ rs;:“ "- "■ ■- 

■ ESI-Si.'; ‘ 


■ ::;s:".,‘i,"‘;,.;:S-:L: 

■ ES'iSE.,;; .5“’"- 

■ “ =~- 1”^ 



lelng a aenber of 

a Navy ftually, he was eonecantly on the Dova and 

actendod eight dll 

‘ferent grade eohools and t-o high schools in Ha-aii, 

Virginia. Callfort 

lia and Eeotuchy. lie studied ceo years at Texas ASH 

IBivoreity and tli. 

■n t-o at the Pennsylvania State University -here he 

received a g.S. in 

1 Aerospace Engineering in 1971, his next three years 

-ere epoat -orhinf 

1 -ich the Notional Acodcay of Scioneo and the Aaiericon 

FsjarholagUal AsBociacloa in Washinyton, D.C. In Sepceeber 1974 he began 
hU graduate education In Fnvlronieantal Engineering Scionces at the 

University of Flor 

■ida. After receiving a Ibster of Engineering in 

August of 1975, he 

stayed on at the university as a graduate research 

assistant in pursu 

it of a Ph.D. for three years, the result ef vhleh is 


Ibis dissertation sas subisittod to tha Gmduota Faculty of the College 
of En|lncerins and to die Graduate Council, end wes accepted as partial 
fulfillioent of the roc[ulre«ent5 for the degree of Doctor of Philosophy.